I'm sorry, I read your reply too hastily. You are talking about adjusting weights for the restraints we already use, not dihedral restraints. By improving the structure you will improve the Rama score, and not the other way around. Outliers flag their residues for inspection in case of errors such as pep-flips or mis-built side chains, but if there is no error they may be real outliers or the data may not be strong enough to keep them from drifting slightly over the contour line; in either case they should not be "fixed" before deposition.
On 02/25/2015 03:36 PM, Robbie Joosten wrote:
Hi Ed, I'm not saying you should restrain to Ramachandran torsion angles; neither Refmac nor pdb_redo does that. Other restraints influence the Ramachandran plot immensely and you should optimize the weights on those (to optimize the free likelihood, not some Ramachandran score). The typical distribution of phi and psi angles are brought on by steric hindrance, so the Van der Waals restraints in the presence of riding hydrogens do most of the work. Also bond length and angle restraints are important: you cannot get a proper backbone conformation when these restraints are too tight. Too loose restraints cause other problems that make your structures model less probable. Cheers, Robbie Sent with my Windows Phone ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Van: Edward A. Berry <mailto:ber...@upstate.edu> Verzonden: 25-2-2015 18:36 Aan: CCP4BB@JISCMAIL.AC.UK <mailto:CCP4BB@JISCMAIL.AC.UK> Onderwerp: Re: [ccp4bb] adjusting bad Ramachandran angles If you use restraints to fix outliers, I strongly suggest to refine to convergence without restraints after they are "fixed". If some outliers return, and "% favored" decreases, so be it. For one thing, depositing a structure with dihedrals restrained gives an unfair impression of higher quality compared to similar structures whose "% favored" actually represents the ability of the data to give the correct conformation. More importantly, you may actually be making a worse model of those (~0.2%?) of residues that are real outliers, and (~5%?) that are really "allowed" and not "favored". At least that's how it used to be: On 12/20/2006 10:34 PM, Gerard DVD Kleywegt wrote: > you have to ask yourself if you really, really, really want to do that. and if the answer is 'yes' you have to ask yourself again > > the reason is that the conformational torsion angles are eminently suited for validation purposes, but only when they are unrestrained. in other words, once you restrain phi and psi, you can't use ramachandran plots any longer (well, you can, but they are useless as validation tools) On 02/25/2015 12:21 PM, Robbie Joosten wrote: > Hi Michael, > > This depends a bit on how bad your backbone torsion angles are and why. Refining with good restraint weights and flipping the odd peptide can help a lot. If you have really low resolution you might need specific restraints. You can try the PDB_REDO webserver at http://xtal.nki.nl/PDB_REDO which will optimise the restraint weights for REFMAC and flip peptides if the density and the Ramachandran plot show that this is needed. You can also try adding hydrogen bond restrains from ProSMART to help the Ramachandran plot in regions that have reasonable secondary structure. This works quite well at resolutions lower than ~3A. > > HTH, > Robbie > >> -----Original Message----- >> From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of >> Michael Murphy >> Sent: Wednesday, February 25, 2015 18:09 >> To: CCP4BB@JISCMAIL.AC.UK >> Subject: [ccp4bb] adjusting bad Ramachandran angles >> >> Does anyone know of a way to adjust Ramachandran angles so that they fall >> within the preferred range? Either in Coot or possibly some online server? I >> have been trying to do it manually without much success, I was wondering >> whether there might another way to do it. -Thanks >
